Wu, Shengli, Li, Dan, Xing, Wenting and Liu, Ying  ORCID: https://orcid.org/0000-0001-9319-5940
2025.
Quantitative prediction model for lithium-ion battery life uncertainty based on DAE-CNN-BiGRU quantile regression.
Journal of Energy Storage
123
, 116771.
10.1016/j.est.2025.116771
|
Preview |
PDF
- Accepted Post-Print Version
Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (2MB) | Preview |
Abstract
To ensure the safety and reliability of lithium-ion battery management systems (BMS), accurately predicting the remaining useful life (RUL) is essential. However, during the operation of lithium-ion batteries, various uncertainties, including energy regeneration and localized fluctuations, introduce significant challenges, making it difficult to predict RUL with the desired accuracy. In this paper, we develop a quantitative model for predicting the uncertainty in the remaining life of lithium-ion batteries. To be specific, the approach begins by employing a denoising auto-encoder (DAE) to reconstruct the original signal during data preprocessing. Next, a one-dimensional convolutional neural network (1D-CNN) is utilized to deeply analyze the capacity data of the lithium-ion batteries. The representative features extracted by the CNN are then fed into a bidirectional gated recurrent unit (BiGRU) network. A quantile regression (QR) layer is integrated into the BiGRU architecture to generate the final predictions of the battery's remaining service life. The quantile regression loss function is applied during the network training process to enhance the accuracy of the remaining service life predictions. Performance evaluation was conducted using publicly available datasets from NASA and CALCE, with comparisons against other prediction methods. Experimental results indicate that the quantile regression approach enhances the accuracy of the gated recurrent unit (GRU) neural network, demonstrating superior predictive performance.
| Item Type: | Article |
|---|---|
| Date Type: | Publication |
| Status: | Published |
| Schools: | Schools > Engineering |
| Publisher: | Elsevier |
| ISSN: | 2352-152X |
| Funders: | Chongqing Natural Science Foundation |
| Date of First Compliant Deposit: | 7 May 2025 |
| Date of Acceptance: | 22 April 2025 |
| Last Modified: | 29 May 2025 13:45 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/178104 |
Actions (repository staff only)
 |
Edit Item |
Altmetric
Altmetric